Commodity information pushing system and method based on precision marketing

Abstract

The invention provides a commodity information pushing system and method based on precision marketing, and relates to the technical field of electronic commerce and information, and the method comprises the steps: collecting multi-source flight operation data and GIS data through a branch and bound method; performing field importance analysis and feature engineering analysis on the data, and selecting feature fields pushed to the commodity information to obtain feature data; according to the feature data, performing deep mining evaluation analysis on the flight operation data to construct a flight operation guarantee capability evaluation model and a sliding optimal path algorithm model; and according to the operation guarantee capability evaluation model, the sliding optimal path algorithm model and the historical behavior data of the user, constructing an accurate portrait of the target user. Through personalized user portrait construction and matching and real-time accurate pushing and marketing effect evaluation, the commodity information pushing accuracy and the user satisfaction degree are improved.

Description

Technical Field

[0001] This invention relates to the fields of e-commerce and information technology, and in particular to a product information push system and method based on precision marketing. Background Technology

[0002] In today's digital and information-driven era, precision marketing has become a crucial means for enterprises to enhance market competitiveness and strengthen customer loyalty. Particularly in the aviation sector, with the continuous increase in the number of flights and the increasing diversification of passenger demands, how to accurately deliver product information tailored to passengers' personalized needs and preferences has become a pressing issue for airlines and related service providers. Traditional product information delivery methods often lack targeting and fail to meet passengers' personalized needs, resulting in poor delivery effectiveness and low user satisfaction. Existing flight operation support capability assessment models are often overly simplistic and unable to accurately assess the overall capacity of flight operations, leading to a lack of scientific basis for subsequent product information delivery.

[0003] Existing product information push methods often lack real-time performance and accuracy, making it difficult to deliver product information to target users at the right time and place. The lack of an evaluation system for the effectiveness of precise pushes makes it difficult to measure the effectiveness of push strategies and user satisfaction, hindering continuous optimization of these strategies. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a product information push system and method based on precision marketing, which improves the accuracy of product information push and user satisfaction through personalized user profile construction and matching, as well as real-time precision push and marketing effect evaluation.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: Firstly, a method for pushing product information based on precision marketing, the method comprising: Multi-source flight operation data and GIS data were collected using the branch-and-bound method. Field importance analysis and feature engineering analysis were performed on the data to select the feature fields for pushing product information and obtain feature data. Based on the feature data, in-depth mining and evaluation analysis of flight operation data are conducted to construct a flight operation support capability assessment model and a taxiing optimal path algorithm model; Based on the operational support capability assessment model, the optimal taxiing path algorithm model, and historical user behavior data, a precise profile of the target user is constructed. The user profile is matched with the product attributes in the product information database to obtain the matching product information; Based on flight status and user real-time location, NSGA-II pushes matched product information to target users to achieve precise marketing of product information.

[0006] Furthermore, multi-source flight operation data and GIS data are collected using the branch-and-bound method, including: The task of collecting multi-source flight operation data and GIS data was defined. The multi-source flight operation data includes AODB, ACDM, ground service support operations, apron operation management, apron vehicle management, and integrated command. The design framework breaks down complex data collection tasks into multiple subtasks, each corresponding to the collection of a specific data source or data type, in order to achieve data collection.

[0007] Furthermore, field importance analysis and feature engineering analysis were performed on the data to select characteristic fields for product information push, resulting in feature data, including: A comprehensive exploratory analysis of the collected data was conducted using statistical methods to assess the impact of each field on the target variable and obtain the analysis results. Based on the analysis results, fields that have an impact on product information push are selected to achieve field importance analysis; Based on the field importance analysis, the feature fields are engineered to obtain feature data, which includes feature selection, feature extraction, and feature transformation. By checking the null value rate of each feature data, feature fields with a null value rate exceeding a set threshold are excluded to obtain the final feature data.

[0008] Furthermore, based on the feature data, in-depth mining and evaluation analysis of flight operation data are conducted to construct a flight operation support capability assessment model and a taxiing optimal path algorithm model, including: Based on the characteristic data, in-depth mining and evaluation analysis of flight operation data are conducted to obtain analysis results; Based on the analysis results, a flight operation support capability assessment model was constructed. Based on the flight operation support capability assessment model, a framework for the taxiing optimal path algorithm model is designed, and a rule-based algorithm is implemented on the algorithm model framework to determine the runway selection and pushback time of the aircraft. Based on the aircraft's runway selection and pushback time, the aircraft's taxiing path is optimized using the particle swarm optimization algorithm to obtain the optimized taxiing path; By integrating the flight operation support capability assessment model and the taxiing optimal path algorithm model, we obtain the flight operation support capability assessment model and the taxiing optimal path algorithm model.

[0009] Furthermore, based on the operational support capability assessment model, the optimal taxiing path algorithm model, and historical user behavior data, a precise profile of the target user is constructed, including: Collect users' historical flight selection preferences and consumption habits data to analyze the common characteristics and differences among different user groups; Based on the common characteristics and differences of different user groups, and combined with the results of the operation and maintenance capability assessment model and the optimal taxiing path algorithm model, a precise profile of the target user is constructed.

[0010] Furthermore, the user profile is matched with the product attributes in the product information database to obtain the matched product information, including: Based on the precise profiles of the target users, a product information database is established; The constructed user profile is matched with the product attributes in the product information database. By calculating the similarity between the user profile and the product attributes, the matched product information is obtained.

[0011] Furthermore, based on flight operation status and user real-time location, matching product information is pushed to target users via NSGA-II to achieve precise marketing product information delivery, including: Based on flight operation status and user's real-time location, obtain the product information matched to the user; Determine optimization goals based on the product information matched to the user; A random initial set of solutions is generated as a population. Each solution represents a matching scheme between a user profile and product attributes, including information such as the matched product ID, push channel, and push time. Each solution in the population is sorted in a non-dominated manner and assigned to different non-dominated levels. Calculate the crowding distance of each solution in the same non-dominated hierarchy, and select parent individuals for crossover based on the non-dominated hierarchy and crowding distance to generate offspring individuals. Perform mutation operations on the offspring individuals to introduce new genetic information. The parent and child generations are merged to form a new population. Non-dominated sorting and crowding distance are calculated. When the maximum number of iterations is reached, the iteration is terminated to obtain the final solution. Based on the final solution, a matching scheme is obtained, and product information is pushed to target users to achieve precise marketing of product information.

[0012] Secondly, a product information push system based on precision marketing includes: The acquisition module is used to collect multi-source flight operation data and GIS data using the branch and bound method; The processing module is used to perform field importance analysis and feature engineering analysis on the data, select the feature fields for pushing product information, and obtain feature data; based on the feature data, it performs in-depth mining and evaluation analysis on flight operation data to build a flight operation support capability assessment model and a taxiing optimal path algorithm model; based on the operation support capability assessment model, the taxiing optimal path algorithm model, and user historical behavior data, it constructs a precise profile of the target user. The matching module is used to match user profiles with product attributes in the product information database to obtain matching product information; based on flight operation status and user real-time location, the matching product information is pushed to target users through NSGA-II to achieve precise marketing product information push.

[0013] Thirdly, a computing device, comprising: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0014] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0015] The above-described solution of the present invention has at least the following beneficial effects: By collecting multi-source flight operation data and GIS data using the branch-and-bound method, the comprehensiveness and accuracy of the data can be ensured. The design framework decomposes the complex data collection task into multiple sub-tasks, improving the efficiency of data collection and reducing its difficulty and cost. A comprehensive exploratory analysis of the data using statistical methods assesses the impact of each field on the target variable, enabling the identification of fields with key influence on product information delivery. The feature fields are engineered, and a flight operation support capability assessment model is built based on the feature data. This model accurately assesses the overall capacity of flight operations. Combining the results of the flight operation support capability assessment model, a framework for a taxiing optimal path algorithm model is designed, and a rule-based algorithm is implemented to determine the aircraft's runway selection and pushback time. Historical flight selection preferences and consumption habits of users are collected, and the common characteristics and differences among different user groups are analyzed. Combining the results of the flight operation support capability assessment model and the taxiing optimal path algorithm model, a precise user profile is constructed. This user profile is matched with product attributes in the product information database to obtain matched product information. This matched product information is then pushed to the target users. Through precise product information delivery, user purchase conversion rates and satisfaction can be improved, enhancing user trust and loyalty to airlines or related institutions. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a product information push method based on precision marketing, provided by an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of a product information push system based on precision marketing, provided by an embodiment of the present invention. Detailed Implementation

[0018] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0019] like Figure 1 As shown, an embodiment of the present invention proposes a product information push method based on precision marketing, the method comprising the following steps: Step 11: Collect multi-source flight operation data and GIS data using the branch and bound method; Step 12: Perform field importance analysis and feature engineering analysis on the data, select the feature fields for pushing product information, and obtain feature data; Step 13: Based on the feature data, conduct in-depth mining and evaluation analysis of flight operation data to construct a flight operation support capability evaluation model and a taxiing optimal path algorithm model; Step 14: Based on the operational support capability assessment model, the optimal taxiing path algorithm model, and historical user behavior data, construct a precise profile of the target user; Step 15: Match the user profile with the product attributes in the product information database to obtain the matched product information; Step 16: Based on the flight operation status and the user's real-time location, push the matched product information to the target user through NSGA-II to achieve precise marketing product information push.

[0020] In this embodiment of the invention, collecting multi-source flight operation data and GIS data using the branch and bound method ensures the rapid location of key information within a massive dataset, improving the efficiency and accuracy of data collection. Utilizing the branch and bound method for field importance analysis and feature selection effectively identifies the feature fields most influential on product information delivery, thereby enhancing the effectiveness of feature engineering. Based on precise feature data, more accurate flight operation support capability assessment models and taxiing optimal path algorithm models can be constructed. These optimized models more accurately reflect the actual situation of flight operations. Based on the operation support capability assessment model and taxiing optimal path algorithm model, combined with user historical behavior data and flight operation data, a more detailed and comprehensive user profile can be constructed. This refined user profile helps to understand user needs and behavioral habits more deeply, laying a solid foundation for precision marketing. Through the NSGA-II optimization algorithm, matching product information is precisely pushed to target users based on flight operation status and real-time user location. This personalized push method significantly improves the acceptance and conversion rate of marketing information, thereby enhancing marketing effectiveness. This demonstrates the implementation of a precision marketing product information push method.

[0021] In a preferred embodiment of the present invention, step 11 above may include: Step 111: Determine the tasks for collecting multi-source flight operation data and GIS data. The multi-source flight operation data includes AODB, ACDM, ground service support operations, apron operation management, apron vehicle management, and integrated command. Step 112: Design a framework to break down the complex data collection task into multiple sub-tasks. Each sub-task corresponds to the collection of a specific data source or data type in order to achieve data collection.

[0022] In this embodiment of the invention, by collecting multi-source flight operation data from airport operation databases, airport collaborative decision-making management systems, ground service support operations, apron operation management, apron vehicle management, and integrated command, comprehensive information on the entire process of flight planning and execution can be achieved. The cross-verification of multi-source data reduces potential errors or biases from single data sources, improving data accuracy and reliability. Combined with GIS data, flight operation information can be integrated with geospatial information to achieve visualized display and analysis of flight operations, providing intuitive support for flight route planning and resource allocation. The complex data collection task is decomposed into multiple sub-tasks, each corresponding to the collection of a specific data source or data type. Parallel processing can be achieved, improving data collection efficiency. Each sub-task has a clearly defined responsible person and task assignment, helping to ensure the smooth and timely completion of data collection.

[0023] The specific steps of this invention include: Step 111: Based on the goals of precision marketing, determine the multi-source flight operation data that needs to be collected, including data from airport operation databases, airline information display systems, ground service support operations, apron operation management, apron vehicle management, and integrated command systems. Identify the GIS data to be collected; list the specific data items for each data source, such as flight number, departure time, arrival time, gate information, and vehicle dispatch information. Based on the data sources and data items, assign data collection tasks to the relevant departments or personnel.

[0024] Step 112 involves a detailed analysis of the data collection tasks identified in Step 111, identifying their complexity and interrelationships. Based on the analysis results, a hierarchical data collection framework is designed. The top layer of the framework represents the overall data collection task, while the bottom layer comprises specific sub-tasks. Each sub-task corresponds to the collection of a specific data source or data type. The overall data collection task is decomposed into multiple sub-tasks, each with clearly defined data source, data items, collection methods, and time requirements. Based on the designed data collection framework and the decomposed sub-tasks, the collected data undergoes preliminary quality checks and verification. The collected data from each sub-task is then integrated to form complete multi-source flight operation data and GIS data.

[0025] In a preferred embodiment of the present invention, step 12 above may include: Step 121: Conduct a comprehensive exploratory analysis of the collected data using statistical methods to assess the impact of each field on the target variable and obtain the analysis results; Step 122: Based on the analysis results, filter out the fields that have an impact on product information push to achieve field importance analysis; Step 123: Based on the field importance analysis, perform engineering processing on the feature fields to obtain feature data, which includes feature selection, feature extraction, and feature transformation. Step 124: By checking the null value rate of each feature data, feature fields with a null value rate exceeding a set threshold are excluded to obtain the final feature data.

[0026] In this embodiment of the invention, exploratory analysis can comprehensively understand the distribution, trends, and outliers of the data, assess the impact of each field on the target variable, and help identify which fields play a key role in product information push, providing a basis for field selection. Exploratory analysis may reveal potential patterns and correlations in the data, providing new ideas for formulating product information push strategies. By analyzing the integrity and consistency of the data, problems in the data can be discovered and addressed, improving data quality. Filtering out fields that have no or minimal impact on product information push can reduce data dimensionality. Field selection allows for more rational allocation of computing and storage resources, improving resource utilization efficiency. Feature selection, extraction, and transformation can yield more representative and stable feature data, improving the model's learning and generalization abilities. Feature engineering helps reduce the impact of noise and outliers in the data on the model. Different models may have different feature requirements; feature engineering can generate feature data suitable for different models, improving model applicability. Reasonable feature engineering can reduce data redundancy and complexity, improve computational efficiency, and reduce training time. Excluding feature fields with excessively high null values ​​ensures data integrity and consistency, preventing null values ​​from impacting model training. Complete feature data helps improve model accuracy, making the push strategy more precise and reliable. Excluding feature fields with excessively high null values ​​simplifies the data processing flow, reducing unnecessary calculations and processing steps.

[0027] The specific steps of this invention include: Step 121: Summarize all collected multi-source flight operation data and GIS data, clean the data, and handle outliers and missing values; based on the data type and analysis objectives, select the mean statistical method, Pearson correlation coefficient correlation analysis, and histogram distribution analysis to perform descriptive statistics on each field, obtain its basic distribution and characteristics, calculate the correlation between the field and the target variable, assess the degree of influence of the field on the target variable, draw distribution maps and correlation matrices, and record the analysis results in the report, including descriptive statistical results, correlation coefficients, distribution maps, etc. for each field.

[0028] Step 122: Based on the analysis objectives, set the criteria for field filtering, such as the threshold for the relevance coefficient; based on the filtering criteria, filter out the fields that have a significant impact on product information push from the analysis results, and for fields with a relevance coefficient higher than the threshold.

[0029] Step 123: From the selected fields, recursively eliminate features to select the most valuable features for model construction. Transform or combine the original features to generate new features; for example, extract hour and minute features from timestamps; extract keywords and word frequencies from text data. Standardize and normalize the features, and integrate the selected, extracted, and transformed features into a feature dataset.

[0030] Step 124: For each feature in the feature dataset, calculate its null value rate. The null value rate is the proportion of missing values ​​to the total number of samples. Based on data quality and model requirements, set a threshold for the null value rate. For example, set the null value rate threshold to 5%, meaning features with a null value rate exceeding 5% will be excluded. Based on the calculated null value rate and the set threshold, exclude feature fields with null value rates exceeding the threshold. The remaining feature fields are retained as the final feature data.

[0031] In a preferred embodiment of the present invention, step 13 above may include: Step 131: Based on the feature data, conduct in-depth mining and evaluation analysis of the flight operation data to obtain the analysis results; Step 132: Based on the analysis results, construct a flight operation support capability assessment model; Step 133: Based on the flight operation support capability assessment model, design the framework of the taxiing optimal path algorithm model, and implement the rule algorithm on the algorithm model framework to determine the runway selection and pushback time of the aircraft. Step 134: Based on the runway selection and pushback time of the aircraft, the taxiing path of the aircraft is optimized using the particle swarm optimization algorithm to obtain the optimized taxiing path; Step 135: Integrate the flight operation support capability assessment model and the taxiing optimal path algorithm model to obtain the flight operation support capability assessment model and the taxiing optimal path algorithm model.

[0032] In this embodiment of the invention, valuable insights can be gained through in-depth mining and analysis of flight operation data. The analysis results help identify potential problems and risks in flight operations, and the construction of an evaluation model can quantitatively assess flight operation support capabilities, including runway capacity, taxiing efficiency, and ground service support. The evaluation model can identify resource bottlenecks and redundancies, thereby optimizing resource allocation and improving resource utilization efficiency. By designing an optimal taxiing path algorithm model, the aircraft's taxiing path can be optimized, reducing taxiing time and distance and improving taxiing efficiency. Rule-based algorithms can ensure the rationality of aircraft runway selection and pushback time, reducing runway conflicts and delays, and improving runway utilization efficiency. By optimizing taxiing paths and runway selection, the coordination between flight operations can be enhanced, improving overall operational efficiency. Particle swarm optimization (PSO) is an intelligent optimization algorithm that can automatically search for optimal solutions to obtain more optimized taxiing paths. PSO has strong adaptability and robustness, and can adapt to different airport environments and flight operation conditions.

[0033] The specific steps of this invention include: Step 131: Based on the feature data, correlate the feature data with flight operation data. According to the analysis objectives, use cluster analysis to identify peak and off-peak periods in flight operations, and use association rule mining to discover the correlation between flight delays and factors such as weather and airport traffic. Use the selected mining analysis methods to conduct in-depth analysis of the flight operation data. The analysis results include discovered patterns, trends, and association rules.

[0034] Step 132: Based on the analysis objectives and data characteristics, a random forest learning model is selected to construct a flight operation support capability assessment model. The model is trained using the feature data and target variable from the analysis results. The target variable is the flight operation support capability index. Model parameters are optimized through methods such as cross-validation and grid search to improve model performance. The trained model is evaluated using a test dataset, and the model's accuracy is calculated. The trained model is then deployed to the production environment for real-time evaluation of flight operation support capabilities.

[0035] Step 133: Based on the results of the flight operation support capability assessment model, design the framework of the taxiing optimal path algorithm model. The framework should include modules such as runway selection, pushback time determination, and taxiing path planning. On the algorithm model framework, implement rule-based algorithms to determine the runway selection and pushback time of aircraft. The rules can be formulated based on factors such as flight type, airport traffic, and runway usage. For example, for large flights, long runways are preferred; for peak periods, pushback time is delayed to avoid congestion.

[0036] Step 134: Initialize the parameters of the particle swarm optimization algorithm, such as the number of particles, number of iterations, and inertia weight; randomly generate an initial particle swarm, with each particle representing a possible gliding path. Evaluate the quality of each gliding path based on gliding time and distance, and optimize the gliding path using the particle swarm optimization algorithm. In each iteration, update the particle position and velocity, evaluate the particle quality based on gliding time and distance, and record the optimal gliding path.

[0037] Step 135: Integrate the flight operation support capability assessment model and the taxiing optimal path algorithm model to form a complete model system.

[0038] In a preferred embodiment of the present invention, step 14 above may include: Step 141: Collect users' historical flight selection preferences and consumption habits data to analyze the common characteristics and differences among different user groups; Step 142: Based on the common characteristics and differences of different user groups, and combined with the results of the operation support capability assessment model and the optimal gliding path algorithm model, construct a precise profile of the target user.

[0039] In this embodiment of the invention, by collecting users' historical flight selection preferences and consumption habits data, a deeper understanding of users' preferences regarding flight time, price, and services can be gained. Analyzing the common characteristics and differences among different user groups helps identify the specific needs and behavioral patterns of each group. By analyzing the differences among user groups, resources can be allocated more rationally, such as providing different service packages or preferential strategies for different user groups, thereby improving resource utilization efficiency. Building precise profiles of target users allows for a more accurate understanding of their needs and preferences. For example, suitable flights, seats, and catering services can be recommended based on user preferences. Precise profiles help to develop more precise marketing strategies and improve marketing effectiveness. Combining the results of the operational support capability assessment model and the taxiing optimal path algorithm model allows for more scientific scheduling of flight plans and resource allocation, improving the efficiency and safety of flight operations. Furthermore, adjusting service strategies based on user profiles can further enhance the overall performance of flight operations.

[0040] The specific steps of this invention include: Step 141: Extract users' historical flight selection records from the airline's database or user management system, including flight number, departure and arrival times, cabin class, and ticketing channel; collect users' consumption records, such as additional service purchases, frequent flyer program participation, and points redemption records; integrate users' basic information, such as age, gender, occupation, and place of residence; and use cluster analysis and association rule mining to identify common characteristics of different user groups in flight selection and consumption habits. Compare the differences between different user groups using analysis of variance to identify key characteristics that distinguish each group. Based on the results of common characteristics and difference analysis, use the K-means clustering algorithm to divide users into different groups, name each user group, and describe its main characteristics and behavioral patterns.

[0041] Step 142: Integrate the user group characteristic data obtained in Step 141 with the results of the flight operation support capability assessment model and the taxiing optimal path algorithm model; define the profile dimensions, where the basic attributes include the user's basic information and flight selection preferences; spending power includes assessing the user's spending power and spending tendency based on the user's spending records; travel needs include the user's time sensitivity and comfort requirements.

[0042] By leveraging the results of the optimal gliding path algorithm model, we infer users' preferences or potential needs regarding gliding paths. Specific tags are defined for each user profile dimension, such as "high spending power" and "time-sensitive user." Weights are assigned based on the importance of each dimension to the user profile. Data mining and neural networks are used to map user data onto profile tags, generating precise user profiles. These precise profiles are then applied to targeted marketing product information pushes, improving push effectiveness and user satisfaction.

[0043] In a preferred embodiment of the present invention, step 15 may include: Step 151: Based on the accurate profiles of the target users, establish a product information database; Step 152: Match the constructed user profile with the product attributes in the product information database. By calculating the similarity between the user profile and the product attributes, the matched product information is obtained.

[0044] In this embodiment of the invention, by accurately profiling target users, their specific needs and preferences for products can be clearly identified, thereby enabling the targeted establishment of a product information database to ensure that the products in the database meet user expectations. Establishing a product information database based on user profiles helps airlines or merchants more accurately predict product demand, thereby optimizing inventory management and reducing inventory backlogs and stockouts. Products in the database better match user needs and preferences, making it easier for users to find their desired products when browsing and purchasing, thus improving user satisfaction. By calculating the similarity between user profiles and product attributes, highly matched product information can be obtained, enabling personalized recommendations and improving the accuracy and effectiveness of recommendations. Personalized recommendations provide users with product information that better meets their needs and preferences, reducing the time users spend sifting through a vast amount of products and enhancing user experience. Matched product information is more likely to attract user attention, stimulate their desire to buy, thereby increasing sales opportunities and sales revenue. By analyzing the matching results between user profiles and product attributes, user purchasing behavior and preferences can be understood, providing data support for optimizing marketing strategies.

[0045] The specific steps of this invention include: Step 151: Collect product information provided by airlines or relevant partner merchants, including flight packages, upgrade services, extra baggage allowance, airport lounge services, and car rental services. Clean and organize the collected product information, removing duplicate, erroneous, or incomplete data, and standardize the format and coding method of the product information. Classify the product information according to the type of product, service content, or target user group. For example, products can be divided into categories such as "flight-related," "extra services," and "travel packages." Store the organized product information in a database to form a product information database, and update the product information database regularly, adjusting the product information and classification methods according to market changes and user feedback.

[0046] Step 152: Analyze the various dimensions and tags in the user profile and map them to the corresponding attributes in the product information database. For example, "high spending power" in the user profile can be mapped to the "price" attribute in the product information database. Based on the characteristics of the user profile and product attributes, select the Euclidean distance similarity calculation method. For text or categorical data, use the TF-IDF method for vectorization before calculating the similarity. For each user profile, calculate its similarity with each product in the product information database using the Pearson correlation coefficient. Based on the similarity calculation results, set a threshold to filter matching product information. Products with similarity higher than the threshold are considered to match the user profile. The matched product information is sorted according to similarity, with products with high similarity being displayed first. The accuracy and effectiveness of the matching results are verified through user surveys, A / B testing, and other methods.

[0047] In a preferred embodiment of the present invention, step 16 above may include: Step 161: Obtain the product information matched to the user based on the flight operation status and the user's real-time location; Step 162: Determine the optimization goals based on the product information matched to the user; Step 163: Randomly generate a set of initial solutions as a population. Each solution represents a matching scheme between a user profile and product attributes, including information such as the matched product ID, push channel, and push time. Perform non-dominated sorting on each solution in the population and assign them to different non-dominated levels. Step 164, through Calculate the crowding distance for each solution within the same non-dominated hierarchy, and select parent individuals for crossover based on the non-dominated hierarchy and crowding distance to generate offspring individuals. Then, perform mutation operations on the offspring individuals to introduce new genetic information. Representing the solution Crowded distance; Indicates the number of objective functions; These are the weights of the m-th objective function; ( ) is the first The objective function value ( A certain transformation function of ). ( ) is the first The maximum value of each objective function and minimum value A certain transformation function is performed; Describe the solutions respectively In the current non-dominated hierarchy, by number The objective function values ​​of adjacent solutions after sorting the objective function values; These represent the maximum and minimum values ​​of all solutions in the current non-dominated hierarchy on the m-th objective function, respectively. It is the adjustment factor in the Sigmoid function, which controls the steepness of the Sigmoid function; It is the center point in the Sigmoid function, used to adjust the center point position of the local density; Step 165: Merge the parent and child generations to form a new population, perform non-dominated sorting and calculate crowding distance. When the maximum number of iterations is reached, terminate the iteration to obtain the final solution. Step 166: Based on the final solution, obtain the matching scheme and push the product information to the target users to achieve precise marketing product information push.

[0048] In this embodiment of the invention, by combining flight operation status and user real-time location, the most timely and relevant product information can be obtained, improving the timeliness and accuracy of the push notifications. Based on the user's real-time location and flight status, the user's current needs and preferences can be inferred, thereby pushing more personalized products. The optimization objective is usually quantifiable, which helps to evaluate and adjust the effect during algorithm execution. Randomly generating initial solutions ensures population diversity and avoids getting trapped in local optima. Non-dominated sorting can quickly identify excellent solutions in the population, and the hierarchical structure formed by non-dominated sorting helps to understand the superiority relationship between solutions. The calculation of crowding distance helps to maintain population diversity when selecting parent individuals and avoids premature convergence. Crossover and mutation operations can introduce new genetic information, explore new regions in the solution space, and help find better solutions. Through continuous iteration, the optimal solution can be gradually approached, improving the quality and stability of the solution. Merging parent and offspring to form a new population helps to search for the optimal solution globally and avoid getting trapped in local optima. Based on the matching scheme obtained from the final solution, it can be ensured that product information is accurately pushed to target users, improving marketing effectiveness.

[0049] The specific steps of this invention include: Step 161: Collect real-time flight operation status data, including flight takeoff and landing times, delays, and runway usage. This is achieved through methods such as user mobile phone positioning and gate scanning to obtain the user's real-time location information. The flight operation status and user's real-time location information are then linked to user profiles and product information databases. For example, based on the user's current flight and gate location, relevant product information is filtered. Combining user profiles and product attributes, product information that highly matches the user's current flight operation status and real-time location is selected.

[0050] Step 162: Define the objective function for the multi-objective optimization problem. For example, maximizing user satisfaction can be achieved by evaluating user satisfaction with the pushed products through user feedback or historical data; maximizing marketing effectiveness can be achieved by evaluating marketing effectiveness through metrics such as click-through rate and conversion rate; minimizing push costs can be achieved by considering the cost of push channels, product discount costs, etc.

[0051] Step 163: Randomly generate a set of initial solutions as a population. Each solution represents a matching scheme between a user profile and product attributes, including information such as the matched product ID, push channel, and push time. Perform non-dominated sorting on each solution in the population and assign them to different non-dominated levels. Randomly generate a set of initial solutions, each solution containing information such as the matched product ID, push channel, and push time. Perform non-dominated sorting on each solution in the population and compare the superiority of the solutions according to the objective function value. Assign the solutions to different non-dominated levels, with solutions at lower non-dominated levels being better.

[0052] Step 164: Calculate the crowding distance for each solution within the same non-dominated level using a formula. The crowding distance reflects the local density of solutions in the target space and is used to maintain population diversity. Select parent individuals based on the non-dominated level and crowding distance, prioritizing solutions with lower non-dominated levels. If the non-dominated levels are the same, select solutions with larger crowding distances. Perform a crossover operation on the selected parent individuals to generate offspring individuals. The crossover operation can be performed using single-point crossover, multi-point crossover, or uniform crossover. Perform mutation operations on the offspring individuals to introduce new genetic information.

[0053] Step 165: Merge the parent and child generations to form a new population. Perform non-dominated sorting and crowding distance calculation on the new population. When the maximum number of iterations is reached or other termination conditions are met (such as convergence of the objective function value), terminate the iteration. Select the solution with the lowest non-dominated level and the largest crowding distance from the final population as the final solution.

[0054] Step 166: Based on the final solution, obtain the matching scheme between user profile and product attributes, including the matching product ID, push channel, push time, and other information, and push the matching product information to the target user through the selected push channel.

[0055] like Figure 2 As shown, embodiments of the present invention also provide a product information push system 20 based on precision marketing, comprising: Module 21 is used to collect multi-source flight operation data and GIS data using the branch and bound method; Processing module 22 is used to perform field importance analysis and feature engineering analysis on the data, select feature fields for product information push, and obtain feature data; based on the feature data, it performs in-depth mining and evaluation analysis on flight operation data to construct a flight operation support capability assessment model and a taxiing optimal path algorithm model; based on the operation support capability assessment model, the taxiing optimal path algorithm model, and user historical behavior data, it constructs a precise profile of the target user. The matching module 23 is used to match user profiles with product attributes in the product information database to obtain matched product information; based on flight operation status and user real-time location, the matched product information is pushed to target users through NSGA-II to achieve precise marketing product information push.

[0056] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for pushing product information based on precision marketing, characterized in that, The method includes: Multi-source flight operation data and GIS data were collected using the branch-and-bound method. Field importance analysis and feature engineering analysis were performed on the data to select the feature fields for pushing product information and obtain feature data. Based on the feature data, in-depth mining and evaluation analysis of flight operation data are conducted to construct a flight operation support capability assessment model and a taxiing optimal path algorithm model; Based on the operational support capability assessment model, the optimal taxiing path algorithm model, and historical user behavior data, a precise profile of the target user is constructed. The user profile is matched with the product attributes in the product information database to obtain the matching product information; Based on flight status and user real-time location, NSGA-II pushes matched product information to target users to achieve precise marketing of product information.

2. The product information push method based on precision marketing according to claim 1, characterized in that, Multi-source flight operation data and GIS data were collected using the branch-and-bound method, including: The task of collecting multi-source flight operation data and GIS data was defined. The multi-source flight operation data includes AODB, ACDM, ground service support operations, apron operation management, apron vehicle management, and integrated command. The design framework breaks down complex data collection tasks into multiple subtasks, each corresponding to the collection of a specific data source or data type, in order to achieve data collection.

3. The product information push method based on precision marketing according to claim 1, characterized in that, Field importance analysis and feature engineering analysis were performed on the data to select feature fields for product information push, resulting in feature data, including: A comprehensive exploratory analysis of the collected data was conducted using statistical methods to assess the impact of each field on the target variable and obtain the analysis results. Based on the analysis results, fields that have an impact on product information push are selected to achieve field importance analysis; Based on the field importance analysis, the feature fields are engineered to obtain feature data, which includes feature selection, feature extraction, and feature transformation. By checking the null value rate of each feature data, feature fields with a null value rate exceeding a set threshold are excluded to obtain the final feature data.

4. The product information push method based on precision marketing according to claim 1, characterized in that, Based on feature data, in-depth mining and evaluation analysis of flight operation data are conducted to construct a flight operation support capability assessment model and a taxiing optimal path algorithm model, including: Based on the characteristic data, in-depth mining and evaluation analysis of flight operation data are conducted to obtain analysis results; Based on the analysis results, a flight operation support capability assessment model was constructed. Based on the flight operation support capability assessment model, a framework for the taxiing optimal path algorithm model is designed, and a rule-based algorithm is implemented on the algorithm model framework to determine the runway selection and pushback time of the aircraft. Based on the aircraft's runway selection and pushback time, the aircraft's taxiing path is optimized using the particle swarm optimization algorithm to obtain the optimized taxiing path; By integrating the flight operation support capability assessment model and the taxiing optimal path algorithm model, we obtain the flight operation support capability assessment model and the taxiing optimal path algorithm model.

5. The product information push method based on precision marketing according to claim 1, characterized in that, Based on the operational support capability assessment model, the optimal taxiing path algorithm model, and historical user behavior data, a precise profile of the target user is constructed, including: Collect users' historical flight selection preferences and consumption habits data to analyze the common characteristics and differences among different user groups; Based on the common characteristics and differences of different user groups, and combined with the results of the operation and maintenance capability assessment model and the optimal taxiing path algorithm model, a precise profile of the target user is constructed.

6. The product information push method based on precision marketing according to claim 1, characterized in that... The user profile is matched with the product attributes in the product information database to obtain the matched product information, including: Based on the precise profiles of the target users, a product information database is established; The constructed user profile is matched with the product attributes in the product information database. By calculating the similarity between the user profile and the product attributes, the matched product information is obtained.

7. The product information push method based on precision marketing according to claim 1, characterized in that, Based on flight status and user real-time location, NSGA-II pushes matched product information to target users to achieve precise marketing, including: Based on flight operation status and user's real-time location, obtain the product information matched to the user; Determine optimization goals based on the product information matched to the user; A random initial set of solutions is generated as a population. Each solution represents a matching scheme between a user profile and product attributes, including information such as the matched product ID, push channel, and push time. Each solution in the population is sorted in a non-dominated manner and assigned to different non-dominated levels. Calculate the crowding distance of each solution in the same non-dominated hierarchy, and select parent individuals for crossover based on the non-dominated hierarchy and crowding distance to generate offspring individuals. Perform mutation operations on the offspring individuals to introduce new genetic information. The parent and child generations are merged to form a new population. Non-dominated sorting and crowding distance are calculated. When the maximum number of iterations is reached, the iteration is terminated to obtain the final solution. Based on the final solution, a matching scheme is obtained, and product information is pushed to target users to achieve precise marketing of product information.

8. A product information push system based on precision marketing, the system implementing the method as described in any one of claims 1 to 7, comprising: The acquisition module is used to collect multi-source flight operation data and GIS data using the branch and bound method; The processing module is used to perform field importance analysis and feature engineering analysis on the data, select the feature fields for pushing product information, and obtain feature data; Based on feature data, in-depth mining and evaluation analysis of flight operation data are conducted to construct a flight operation support capability assessment model and a taxiing optimal path algorithm model; based on the operation support capability assessment model, the taxiing optimal path algorithm model, and user historical behavior data, a precise profile of the target user is constructed. The matching module is used to match user profiles with product attributes in the product information database to obtain matching product information; Based on flight status and user real-time location, NSGA-II pushes matched product information to target users to achieve precise marketing of product information.

9. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images